| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Neuroradiology |
1 From the Division of Radiological Sciences, Wake Forest University School of Medicine, Winston-Salem, NC. From the 2001 RSNA scientific assembly. Received January 17, 2002; revision requested February 27; final revision received August 5; accepted August 19. Supported in part by a grant from the Charles A. Dana Foundation. Address correspondence to A.S.F., Department of Radiology, University of Wisconsin Medical School, 600 Highland Ave, Module E3/311, Madison, WI 53792-3252 (e-mail: asfield@wisc.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Twenty adults (16 men, four women; age range, 24-64 years) categorized as low (mean, 41 mg/d) or high (mean, 648 mg/d) caffeine users underwent quantitative flow-sensitive alternating inversion-recovery perfusion magnetic resonance (MR) imaging twice: 90 minutes after a dose of caffeine (250 mg) on one day and after a dose of placebo on another day (randomized counterbalanced design). Doses were preceded by 30 hours of caffeine abstinence to induce withdrawal in high caffeine users. Quantitative CBF maps were gray matter (GM)–white matter (WM) segmented and subjected to region-of-interest analysis to obtain mean CBF in WM, anterior circulation GM (AGM), and posterior circulation GM (PGM). By using two-way repeated-measures analysis of variance, regional CBF data were tested for within-subject differences between caffeine and placebo and for between-subject differences related to dietary caffeine habits. Linear regression was used to determine whether dietary caffeine use predicts CBF or CBF response to caffeine.
RESULTS: Caffeine reduced CBF (P 
.05) by 23% (AGM, PGM) and 18% (WM) in all subjects. Postplacebo (withdrawal) CBF in high caffeine users exceeded that in low users (P .05) by 31% (AGM) and 32% (WM) (PGM, not significant). Mean postcaffeine CBF reduction in AGM was 26% in high users versus 19% in low users (P .05; PGM and WM, not significant). Increasing caffeine consumption predicted higher CBF (P .05) in all regions: r = 0.79 (AGM), 0.57 (PGM), and 0.76 (WM). Dietary caffeine use did not predict CBF response to caffeine.
CONCLUSION: Dietary caffeine consumption and withdrawal are potential confounding variables in cerebral perfusion and functional MR imaging.
© RSNA, 2003
Index terms: Brain, perfusion, 10.12144 • Drugs, side effects • Magnetic resonance (MR), perfusion study, 10.12144
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
The combination of its vasoconstrictive effects and its extreme popularity make caffeine an obvious source of potential error in functional brain imaging experiments, as noted in a recent review (6). However, dietary caffeine consumption remains underrecognized as a confounding variable in the cerebral perfusion imaging literature. In an informal survey of the experimental methods of several recent cerebral perfusion studies, none reported controlling for dietary caffeine effects. (By way of contrast, dietary caffeine is well recognized in cardiac perfusion imaging as responsible for falsely negative dipyridamole–thallium 201 studies [7].) This is especially troublesome, given the recent growth of interest in quantitative perfusion imaging techniques, including dynamic susceptibility contrast magnetic resonance (MR) imaging and arterial spin labeling, and the large number of centers engaged in their development and application. Dietary caffeine consumption may also confound the analysis of brain activation studies with blood oxygen level–dependent (BOLD) contrast, owing to coupling of the BOLD signal to changes in CBF (8,9). This point is underscored by findings in a recent study of caffeine as a potential "BOLD contrast booster" (10,11) and by results from our own still more recent work in this area (12,13).
At first glance, the problem of dietary caffeine and its potentially confounding effects on cerebral perfusion and functional MR experiments would appear to have a simple solution: restricting caffeine intake by experimental subjects, before imaging, for a period equivalent to several caffeine half-lives (the biologic half-life of caffeine in humans ranges from 2.5 to 4.5 hours [14]). Unfortunately, this easily implemented measure might not eliminate the influence of dietary caffeine, owing to the potential for withdrawal effects in habitual caffeine users. Symptoms of a withdrawal syndrome, such as headache, fatigue, and dysphoric mood changes, have been documented after cessation of daily caffeine doses as low as 100 mg (15), the approximate equivalent of just a single cup of coffee (5).
The purpose of our investigation was to evaluate the effects of dietary caffeine intake and withdrawal on CBF, as determined from a randomized, blinded, placebo-controlled study.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Each subject underwent quantitative perfusion MR imaging on two separate days at the same time of day and with matching conditions. However, there was one exception: At 60–90 minutes before imaging, subjects received an oral dose of either caffeine (250 mg, equivalent to two to three cups of coffee) or placebo. Subjects were randomized to receive caffeine on one day and placebo on the other in a single-blind counterbalanced design. Administration of all doses of caffeine or placebo was preceded by at least 30 hours of caffeine abstinence such that the placebo condition would likely reflect a state of withdrawal in the high caffeine users (5). Subjects also underwent functional MR imaging as part of a separate study reported elsewhere.
Imaging Protocol
Experiments were conducted with a 1.5-T MR imaging unit (Echo-speed Horizon LX; GE Medical Systems, Milwaukee, Wis) with a birdcage head coil. T1-weighted sagittal localizer images were acquired with a spin-echo sequence and used to position one transverse section parallel and 10 mm cephalad to the anterior-posterior commissural line. A relative CBF image for this section was obtained by using a flow-sensitive alternating inversion-recovery sequence, which consisted of alternating section-selective and nonselective radio-frequency inversion pulses, a diffusion gradient (equivalent b value of 5.25 mm2/sec) for suppression of intraarterial spins (18), and a single-shot spiral readout gradient (5,100/8/1,600 [repetition time msec/echo time msec/inversion time msec], 7-mm thickness, in-plane resolution of 3.75 x 3.75 mm). A hyperbolic secant pulse was used for the radio-frequency inversion. The width of the section-selective inversion pulse was 20 mm wider than that of the imaging section (10 mm on either side). For field inhomogeneity correction, the first two shots were acquired at slightly different (2 msec) echo times, each after a section-selective radio-frequency inversion pulse. The alternating section-selective and nonselective inversion series were started after the initial two shots and were repeated 31 times to achieve a reasonable compromise between signal-to-noise ratio and imaging time (5 minutes 26 seconds).
The apparent tissue T1 image for the section was calculated by curve-fitting the signal time-course at each voxel following a section-selective inversion-recovery sequence (6,000/8 [repetition time msec/echo time msec]), which was repeated as inversion time was varied from 50 to 4,000 msec in 20 equal increments. A quantitative CBF image was then calculated from the relative CBF and T1-weighted images by using a published perfusion model (18).
Image Processing and Statistical Analysis
Image reconstruction and postprocessing were accomplished at a workstation (UltraSPARC; Sun Microsystems, Santa Clara, Calif) with mathematical software (MATLAB, Mathworks, Sherborn, Mass; IDL, Research Systems, Boulder, Colo). Our procedures for gray matter (GM)–white matter (WM) segmentation and region-of-interest analysis were published previously (19) and will be summarized herein. GM and WM CBF data were processed separately by automatically segmenting the T1-weighted images with brain mapping software (SPM99, Statistical Parametric Mapping; Wellcome Department of Cognitive Neurology, University College London, England) and then applying the resultant masks to the CBF images. The GM was further segmented into two regions corresponding approximately to the territories of the anterior (internal carotid) and posterior (vertebrobasilar) circulations by using semiautomatic region-of-interest analysis software developed in house and supervised by the first author (A.S.F.) in all cases. Anterior circulation GM (AGM) comprised territories of the anterior and middle cerebral arteries, as delineated by the region-of-interest software (19). Mean CBF values were then calculated separately for the WM, AGM, and posterior circulation GM (PGM).
The CBF data were analyzed by using two-way analysis of variance with repeated measures on the treatment factor (caffeine vs placebo). This design tested within-subject differences between caffeine and placebo, between-subject differences related to level of dietary caffeine use, and interaction effects between dietary caffeine use and caffeine-induced CBF response. In addition, since the categorization of subjects into low– and high–caffeine use groups was necessarily somewhat arbitrary, linear regression was used to test whether the subjects’ dietary caffeine habits predicted either their CBF in the placebo condition or the relative magnitude of their CBF response to caffeine. The criterion for statistical significance was chosen as a P value of .05 or less for all tests.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
|
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Caffeine withdrawal syndrome in humans has also been extensively studied and reviewed (5,22). Withdrawal symptoms, with those most commonly reported being headache, fatigue, irritability, and impaired concentration, usually begin approximately 12–24 hours after cessation of caffeine intake and peak after 20–48 hours (22). Importantly, there is considerable evidence to support a relationship between caffeine withdrawal headache and changes in CBF. Mathew and Wilson (21), by using 133Xe inhalation and multiple scintillation detectors, found increased frontal lobe CBF values in high caffeine users after 24 hours of caffeine abstinence. By using transcranial Doppler ultrasonography (US), Couturier et al (23) found increased blood flow velocities in basilar and posterior cerebral arteries after 24 hours of caffeine abstinence, though they did not report the caffeine habits of their subjects. More recently and also by using transcranial Doppler US, Jones et al (24) found increased blood flow velocities in anterior and middle cerebral arteries after 20 hours of caffeine abstinence in subjects who reported moderate dietary caffeine intake (mean, 333 mg/d).
Given the effect of caffeine on CBF known from these prior studies, it was not unexpected to find significantly lower CBF values after administration of a dose of caffeine in the present study. However, two important points from our results warrant emphasis. The first point regards the anatomic distribution of the effect: Whereas researchers in prior studies reported only whole-brain or cortical CBF data, we have shown that the effects of caffeine on CBF pertain to both GM and WM. That this would be the case is not necessarily obvious, given the significant difference between GM and WM in metabolic activity and normal resting CBF and the potential for regional variations in adenosine receptor densities. The second point to be made is that the experimental protocol was designed to simulate conditions that could easily occur in the course of a cerebral perfusion experiment, which potentially could lead to erroneous conclusions. For example, if a subject participating in an experimental study were to ingest, before undergoing quantitative perfusion imaging, two to three cups of coffee (or probably less in the case of the European-style "gourmet" coffee beverages now popular in the United States), the measured CBF values would be falsely low. Further, they would not necessarily be comparable to those of another subject in the same study or to those of the same subject on a different occasion. In this context, "falsely" means that CBF values have been altered by factors extraneous to whatever pathologic condition, treatment, or natural phenomenon is under investigation. A similar caveat applies to BOLD functional MR imaging studies (12,13).
The obvious and easy way to avoid this problem, of course, would be to restrict caffeine intake before perfusion imaging. Unfortunately, the other major finding in our study indicates that this would merely replace the problem of falsely reduced CBF values with one of falsely elevated values, at least in the case of habitual caffeine users and possibly excluding the PGM. The association we observed between higher levels of dietary caffeine intake and higher CBF values in the withdrawal condition was significant by using analysis of variance and linear regression in the AGM and WM. In the PGM, the association was weakest by using regression and not significant by using analysis of variance. These results are similar to those of Mathew and Wilson (21), whose observation of elevated withdrawal CBF values in habitual caffeine users was significant only in the frontal lobes. This apparent difference between anterior and posterior circulations might be explained by regional variations in the numbers of perivascular adrenergic nerves (25), adenosine receptors (25–27), or nucleoside transport proteins (28). However, the PGM regions showed significantly higher intersubject variance than was shown in the AGM and WM regions, which likely occurred as a result of the smaller numbers of voxels contained in the PGM regions, and this finding may have precluded detection of an effect in the PGM by using analysis of variance.
Again, the experimental conditions (30 hours of withdrawal and an average daily intake of 648 mg) that produced these results could easily occur in the course of a cerebral perfusion experiment. According to recent estimates (29), 54% of adults in the United States drink coffee every day, coffee drinkers consume an average of 3.1 cups per day (>500 mg), and 62% of all coffee is consumed at breakfast. Therefore, there is a substantial likelihood that a given subject in a cerebral perfusion experiment, instructed to skip morning coffee before perfusion imaging, would meet the conditions of the current study and manifest erroneous CBF values.
Mathew and Wilson (21) postulated two possibilities for the mechanism by which CBF is increased in caffeine withdrawal: (a) tolerance to the cerebral vasoconstricting effects of caffeine in habitual users, with rebound vasodilatation after its withdrawal, and (b) vasodilatation in response to increased sensory stimulation engendered by the somatic discomfort and pain (headache) of caffeine withdrawal. Although we did not quantify the somatic discomfort of our subjects, the latter possibility is insufficient to explain our observations because the high caffeine users had higher CBF than the low users had, even after receiving a dose of caffeine. On the basis of the data, one may argue, therefore, for an upward CBF "baseline shift" in the high users. Such a shift is consistent with the idea that the brain adapts to chronically high caffeine levels by upregulating its population of adenosine receptors (30,31), leaving a greater number of receptors available to bind endogenous adenosine when caffeine is withdrawn. Regional variations in adenosine receptor populations could explain the differences we observed between the anterior and posterior circulations.
A few methodologic limitations of the present study are worthy of mention. The categorization of subjects into high– and low–caffeine user groups was necessarily somewhat arbitrary and based on the subjects’ own estimates of their dietary caffeine consumption, which could be unreliable. Almost all of the subjects’ caffeine intake came from coffee beverages, which vary widely in their caffeine content. Every effort was made to account for different beverage sizes and method of preparation (eg, instant, drip, espresso). Subjects were also relied on to comply with caffeine restrictions; no verification of compliance was obtained. It is unlikely, however, that any of these limitations would have introduced a systematic bias in the data; rather, they would only serve to increase the variance in the data and dilute the effects that we observed.
The data, though preliminary and obtained from a small number of subjects, implicate dietary caffeine consumption and, in habitual caffeine users, caffeine withdrawal as potential confounding variables in cerebral perfusion experiments. Although the importance of dietary caffeine might be debatable in the case of a single-subject experiment, particularly if the acquired CBF data are neither quantitative nor compared across repeat measurements, a study involving quantitative CBF measurements in multiple subjects or longitudinal measurements in a single subject would certainly demand special efforts to control for effects of caffeine.
It then remains to decide how best to implement such controls. The goals are to maintain unbiased study populations and simultaneously to minimize both intersubject and day-to-day intrasubject variations in cerebral perfusion, which would limit the detection of small changes related to diseases, treatments, or experimental manipulations. One approach would be to exclude regular caffeine users from multisubject cerebral perfusion and functional MR imaging studies and impose caffeine restrictions on subjects before imaging. Of course, this would put a severely limiting constraint on the recruitment of study participants, given the high proportion of regular coffee drinkers among adults worldwide. A less constraining, but still difficult, approach would be to include both coffee drinkers and nondrinkers in multisubject studies and treat caffeine use explicitly as an experimental variable. An investigator who takes this approach must be mindful of the fact that an experimental subject’s "caffeine status," as it pertains to cerebral perfusion, depends on both the subject’s caffeine use habits and the time since the last dose of caffeine was ingested. The reliability of such an approach may be questionable, given the difficulty of quantifying chronic caffeine use accurately and the interindividual variability of the biologic half-life of caffeine. The latter would be easily addressed by measuring plasma caffeine levels at the time of imaging, but the former would be more challenging.
There are two potential compromises that would seem reasonable. One is to impose a monitored period of caffeine abstinence before imaging that would be of sufficient duration to minimize the withdrawal effect. The present study does not point to a specific duration for this period of abstinence except to suggest that it should be considerably longer than 30 hours. The other approach is to have all experimental subjects, including heavy caffeine users, occasional users, and abstainers, adhere to their usual caffeine intake patterns before imaging. This approach relies on the assumption that the human brain regulates its populations of adenosine receptors to reach a new state of equilibrium at whatever caffeine levels are chronically present and that all brains are comparable for purposes of cerebral perfusion imaging when "usual" caffeine levels are maintained. Our data support this assumption in that CBF values were comparable between heavy caffeine users who ingested caffeine and low users who ingested the placebo. This approach might be compromised by subjects with highly erratic caffeine habits but otherwise strikes a reasonable balance between the goals stated previously and the concern for overly limiting recruitment criteria.
Finally, it should be noted that while this study addressed only caffeine use, future studies might address a number of other potential confounders of cerebral perfusion and functional MR imaging experiments, and these confounders would include use of tobacco, alcohol, and over-the-counter and prescription drugs.
In conclusion, the results of this preliminary study implicate dietary caffeine consumption and withdrawal as potential sources of error in the analysis of findings of quantitative cerebral perfusion experiments in particular and hemodynamically based functional brain imaging studies in general. On the basis of the data, one may argue for special efforts on the part of brain imaging researchers to control for the effects of dietary caffeine in future imaging studies. Several possible approaches to implementation of such controls have been discussed in light of the data presented.
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, A.S.F.; study concepts, all authors; study design, D.M.M., A.S.F., P.J.L., J.H.B.; literature research, A.S.F., P.J.L.; clinical studies, A.S.F.; data acquisition and analysis/interpretation, A.S.F., P.J.L., Y.F.Y.; statistical analysis, A.S.F.; manuscript preparation and editing, A.S.F.; manuscript definition of intellectual content, revision/review, and final version approval, all authors.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  A.R. Deibler, J.M. Pollock, R.A. Kraft, H. Tan, J.H. Burdette, and J.A. Maldjian Arterial Spin-Labeling in Routine Clinical Practice, Part 2: Hypoperfusion Patterns AJNR Am. J. Neuroradiol., August 1, 2008; 29(7): 1235 - 1241. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Waaijer, M. S. van Leeuwen, M. J.P. van Osch, B. H. van der Worp, F. L. Moll, R. T. H. Lo, W. P. T. M. Mali, and M. Prokop Changes in Cerebral Perfusion after Revascularization of Symptomatic Carotid Artery Stenosis: CT Measurement Radiology, November 1, 2007; 245(2): 541 - 548. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C Hoecker, M Nelle, B Beedgen, J Rengelshausen, and O Linderkamp Effects of a divided high loading dose of caffeine on circulatory variables in preterm infants Arch. Dis. Child. Fetal Neonatal Ed., January 1, 2006; 91(1): F61 - F64. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Hendrikse, M. J. P. van Osch, D. R. Rutgers, C. J. G. Bakker, L. J. Kappelle, X. Golay, and J. van der Grond Internal Carotid Artery Occlusion Assessed at Pulsed Arterial Spin-labeling Perfusion MR Imaging at Multiple Delay Times Radiology, December 1, 2004; 233(3): 899 - 904. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. J. Laurienti, A. S. Field, J. H. Burdette, J. A. Maldjian, Y.-F. Yen, and D. M. Moody Relationship between Caffeine-Induced Changes in Resting Cerebral Perfusion and Blood Oxygenation Level-Dependent Signal AJNR Am. J. Neuroradiol., September 1, 2003; 24(8): 1607 - 1611. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
